A method for verifying digital energy meters based on AI-based fault simulation testing

Abstract

The application discloses a kind of based on AI's fault simulation analog test digital electric energy meter verification method, steps include: obtaining electric energy meter historical operation data, using the multi-modal fault simulation algorithm based on artificial intelligence, through the analysis electric energy meter historical operation data generates fault mode;Based on the intensity of fault mode, through the electric energy meter performance degradation prediction algorithm based on fault mode analysis the influence of fault mode on electric energy meter performance, the response value of electric energy meter performance parameter is calculated;Based on electric energy meter performance parameter response value, through digital electric energy meter performance verification algorithm, the verification error of each performance parameter is calculated;According to performance parameter verification error, through adaptive parameter updating algorithm optimization parameter.The application has the advantages of comprehensive fault simulation, accurate performance degradation prediction, high verification error sensitivity, good adaptability, fast parameter convergence and the like.

Description

Technical Field

[0001] This invention belongs to the field of electricity meter testing and verification technology, and specifically relates to a method for verifying digital electricity meters. Background Technology

[0002] With the rapid development of smart grids and digital electricity metering technologies, digital energy meters, as core devices for electricity metering, monitoring, and management, have become an important component of modern power systems. Compared to traditional mechanical energy meters, digital energy meters offer higher metering accuracy, more powerful data processing capabilities, more stable communication performance, and a wider range of applications, playing a crucial role in ensuring the stable operation of power systems, guaranteeing the accuracy of metering data, and protecting the rights and interests of electricity users. However, traditional energy meter verification methods are typically time-consuming and labor-intensive, especially during on-site verification, which suffers from low efficiency, susceptibility to human error, or improper equipment debugging. Furthermore, environmental interference may prevent accurate reflection of the meter's actual operating status.

[0003] Therefore, those skilled in the art typically use a verification method based on fault simulation testing to test and verify electricity meters. By simulating different fault scenarios in a simulation environment, the performance of electricity meters under various working conditions is evaluated, thereby verifying their metering accuracy and comprehensively improving the efficiency and accuracy of electricity meter verification.

[0004] However, existing verification methods still have the following problems:

[0005] 1. The complexity and diversity of fault modes are difficult to fully simulate: Existing methods rely on manually preset single fault modes, which cannot fully cover the complex and diverse fault scenarios that electricity meters may encounter in actual operation.

[0006] 2. Insufficient accuracy in performance degradation prediction: Existing methods use linear models, which are difficult to accurately reflect the complex relationship between fault intensity and performance degradation.

[0007] 3. Insufficient sensitivity and adaptability to performance verification errors: Existing methods compare predicted values ​​with standard values ​​using a fixed threshold to calculate static errors, ignoring the impact of fault severity and performance degradation on the errors.

[0008] 4. Insufficient convergence and stability in the parameter optimization process: Parameter adjustment (such as fault amplitude) relies on manual trial and error or simple gradient descent, which has a slow convergence speed and is prone to getting stuck in local extrema or oscillations. Especially in complex fault modes, parameter optimization is inefficient and it is difficult to quickly adapt to diverse operating conditions, resulting in a tradeoff between prediction accuracy and testing efficiency. Summary of the Invention

[0009] This invention proposes an AI-based fault simulation test method for digital energy meter verification. Its purpose is to solve the problems of existing verification methods, such as the inability to fully simulate the complexity and diversity of fault modes, insufficient accuracy in predicting performance degradation, insufficient sensitivity and adaptability of performance verification errors, and insufficient convergence and stability in the parameter optimization process.

[0010] The technical solution of this invention is as follows:

[0011] A fault simulation test method for digital energy meter verification based on AI includes the following steps:

[0012] Step S1: Obtain historical operating data of the electricity meter and use an artificial intelligence-based multimodal fault simulation algorithm to generate fault modes by analyzing the historical operating data of the electricity meter;

[0013] Step S2: Based on the fault mode intensity, analyze the impact of fault modes on the performance of the electricity meter using a fault mode-based electricity meter performance degradation prediction algorithm, and calculate the response values ​​of the electricity meter performance parameters.

[0014] Step S3: Based on the response values ​​of the energy meter performance parameters, calculate the verification error of each performance parameter using a digital energy meter performance verification algorithm;

[0015] Step S4: Verify the error based on the performance parameters, and optimize the parameters using an adaptive parameter update algorithm.

[0016] As a further improvement to the AI-based fault simulation test method for digital energy meter verification, step S1 specifically includes:

[0017] Step S1-1: Obtain historical operating data of the electricity meter. The historical operating data includes the operating records of the electricity meter under various environmental conditions. Divide the historical operating data of the electricity meter into a training set and a validation set. The training set is used to train the artificial intelligence model, and the validation set is used to evaluate the accuracy of the fault modes generated by the artificial intelligence model.

[0018] Step S1-2: Use an AI-based multimodal fault simulation algorithm to generate multiple fault modes by analyzing historical operating data of the electricity meter.

[0019] As a further improvement to the AI-based fault simulation test method for digital energy meter verification, the AI-based multimodal fault simulation algorithm generates fault modes by analyzing historical operating data of the energy meter, combines three fault influencing factors: periodic, transient, and sudden, and introduces random noise to enhance realism, simulating complex fault scenarios that digital energy meters may encounter in actual operation.

[0020] Periodic faults manifest as disturbances that fluctuate regularly over time, and are used to simulate a sine wave-like fluctuation pattern.

[0021] Transient faults are simulated by an exponential decay function to represent rapid changes that occur in a short period of time. The duration and scope of the transient fault can be controlled by adjusting the decay rate.

[0022] Sudden failures are used to simulate unexpected events;

[0023] Random noise is generated according to a normal distribution, and its fluctuation range is controlled by a standard deviation parameter.

[0024] As a further improvement to the AI-based fault simulation test method for digital energy meter verification: in step S2, the fault mode-based energy meter performance degradation prediction algorithm calculates the direct impact on specific performance parameters of the energy meter for each fault mode, introduces a sensitivity coefficient, and measures the sensitivity of a specific fault to a specific performance parameter.

[0025] As a further improvement to the AI-based fault simulation test digital energy meter verification method: in step S2, the energy meter performance degradation prediction algorithm based on fault mode performs nonlinear processing on the fault mode intensity to reflect the differences in nonlinear response of different energy meters to faults.

[0026] As a further improvement to the AI-based fault simulation test digital energy meter verification method: In step S2, the energy meter performance degradation prediction algorithm based on fault mode reflects the fault resistance capability of performance parameters through a dynamic resistance factor. The resistance baseline under fault-free conditions is used as the initial value. As the fault mode intensity increases, the dynamic resistance factor gradually decreases, reflecting the performance degradation of the energy meter under continuous faults. The decay rate of the dynamic resistance factor is controlled by the degradation coefficient.

[0027] As a further improvement to the AI-based fault simulation test method for digital energy meter verification: In step S3, the digital energy meter performance verification algorithm processes each performance parameter one by one, and the processing method is as follows:

[0028] Extract the response values ​​of the performance parameters of the electricity meter and the corresponding standard thresholds, calculate the relative deviation between the two, introduce a pre-set weighting factor to reflect the importance of different performance parameters in the overall performance of the electricity meter, and eliminate the influence of positive and negative signs by squaring to reflect the absolute magnitude of the deviation.

[0029] Based on the cumulative information of failure mode intensity, a dynamic severity factor is calculated to enhance the sensitivity of errors to severe failures.

[0030] A dynamic adjustment term is introduced to adjust the verification error according to the severity of the fault and the degree of performance degradation. Specifically, the reciprocal of the sum of the energy meter performance parameter response value and the smoothing constant is taken and then multiplied by the dynamic severity factor to obtain the amplification term, which amplifies the verification error and makes it more sensitive to capture the negative impact of severe faults on performance.

[0031] The square root of the above calculation results is taken to obtain the final verification error value for each performance parameter.

[0032] As a further improvement to the AI-based fault simulation test method for digital energy meter verification: when calculating the dynamic severity factor, an initial severity coefficient is used as a benchmark value, and then dynamically adjusted according to the ratio of the total intensity of all current fault modes to the preset maximum fault mode intensity.

[0033] As a further improvement to the AI-based fault simulation test method for digital energy meter verification: In step S4, before adjusting the parameters, the direction and degree of the influence of parameter changes on the verification error are determined, i.e., the sensitivity of the parameters is calculated: the sensitivity of the verification error to the fault mode and the sensitivity of the fault mode to the target parameter are analyzed and combined through the chain method to obtain the overall influence of the parameters on the verification error; an exponential decay factor is introduced. When the error is large, the value of the exponential decay factor is small, and the influence of sensitivity is amplified, thereby accelerating the adjustment. When the error is small, the exponential decay factor is close to 1, and the influence of sensitivity is moderately weakened.

[0034] A momentum term is introduced to smooth the parameter update process and reduce oscillations during adjustment. The momentum term is calculated based on the difference between the current parameter value and the previous parameter value, multiplied by a momentum coefficient.

[0035] As a further improvement to the AI-based fault simulation test digital energy meter verification method: in step S4, in order to ensure the physical rationality of the parameters, range constraints are set to prevent the parameters from exceeding the actual acceptable range.

[0036] Compared with the prior art, the present invention has the following advantages:

[0037] 1. This invention acquires historical operating data from electricity meters (such as voltage fluctuations, temperature changes, humidity effects, and external interference) and combines it with an AI-based multimodal fault simulation algorithm to automatically generate composite fault modes that include periodic, transient, and abrupt faults. Based on AI learning, fault modes can be adaptively generated and dynamically adjusted according to the historical operating data of different electricity meters. This avoids the limitations of human assumptions in traditional methods, significantly improving the realism and intelligence of the test, and comprehensively enhancing the coverage of fault modes.

[0038] 2. The energy meter performance degradation prediction algorithm of this invention uses the fault mode intensity as input, introduces a sensitivity coefficient, nonlinear transformation, and dynamic resistance factor, and calculates the response value of energy meter performance parameters (such as metering accuracy). This method intuitively reflects the performance of the energy meter under different fault intensities, accurately captures the cumulative effect of faults on performance and individual differences, and can accurately quantify the impact of faults on energy meter performance, providing a reliable basis for performance degradation analysis. Furthermore, due to the use of nonlinear processing and resistance factor adjustment, this method is applicable to energy meters of different types and states, has stronger adaptability, and significantly reduces the need for one-by-one testing of a large number of samples in traditional manual testing.

[0039] 3. The digital energy meter performance verification algorithm of this invention comprehensively considers the impact of faults and the importance of performance. By introducing dynamic adjustment terms and dynamic severity factors, it amplifies the impact of severe faults on errors, facilitating the timely detection of potential energy meter failure risks and exhibiting good sensitivity and adaptability. Furthermore, the error magnitude is processed by square root extraction, resulting in intuitive and reasonable numerical values ​​that facilitate subsequent parameter adjustments and performance improvements, making the verification results more closely reflect actual application requirements.

[0040] 4. To address the issues of insufficient convergence and stability during parameter optimization, this invention optimizes parameters (such as fault amplitude) through an adaptive parameter update algorithm based on performance parameter verification errors. This makes the fault modes and performance prediction results closer to standard values, improving the reliability of the testing system. By introducing a momentum term and an exponential decay factor, iterative convergence is significantly accelerated, computational resource consumption is reduced, and potential oscillations during optimization are avoided. Furthermore, by setting range constraints, the physical rationality of the parameters is ensured, making the optimization results applicable to real-world electricity meter testing scenarios, thus enhancing the reliability and stability of the testing system. Attached Figure Description

[0041] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0042] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0043] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0044] like Figure 1 A fault simulation test method for digital energy meter verification based on AI includes the following steps:

[0045] Step S1: Obtain historical operating data of the electricity meter and use an artificial intelligence-based multimodal fault simulation algorithm to generate multiple fault modes by analyzing the historical operating data of the electricity meter.

[0046] The specific process is as follows:

[0047] Step S1-1: Obtain historical operating data of the electricity meter, including operating records of the electricity meter under various environmental conditions, such as voltage fluctuations, temperature changes, humidity effects, and external interference (such as electromagnetic interference or power outages). After processing, the historical operating data of the electricity meter is divided into a training set and a validation set. The training set is used to train the artificial intelligence model, and the validation set is used to evaluate the accuracy of the fault modes generated by the artificial intelligence model.

[0048] Step S1-2: Use an AI-based multimodal fault simulation algorithm to generate multiple fault modes by analyzing historical operating data of the electricity meter.

[0049] The core of this algorithm lies in combining periodic, transient, and abrupt faults, while introducing random noise to enhance realism, aiming to simulate the complex fault scenarios that digital energy meters may encounter in actual operation.

[0050] The failure modes include three types of failure influencing factors: periodic failure, transient failure, and sudden failure.

[0051] The periodic fault manifests as a disturbance that fluctuates regularly over time, simulating a sine wave-like fluctuation pattern, such as periodic voltage spikes, with the starting point and fluctuation speed defined by an oscillation frequency and phase shift.

[0052] Based on periodic faults, transient faults are further superimposed. The transient faults are simulated by an exponential decay function to represent rapid changes that occur in a short period of time. The duration and scope of the transient faults are controlled by adjusting the decay rate. When the decay rate is fast, the impact of the fault disappears quickly; when the rate is slow, the impact lasts for a longer period of time. Combined with the decay mechanism of periodic faults, this forms a composite pattern that has both volatility and short-lived impacts.

[0053] The sudden fault simulates a sudden event, such as a momentary power outage. It is represented by a step change that indicates that the fault suddenly occurs and persists at a certain moment. Specifically, when the time reaches a certain preset starting point, the fault intensity will suddenly increase to a fixed value and remain unchanged thereafter. If the current time is less than the starting point, the jump effect is zero. Once the time exceeds the starting point, the jump effect takes effect immediately.

[0054] To enhance the realism and unpredictability of the failure modes, random noise is further introduced. The random noise is generated according to a normal distribution, and the fluctuation range is controlled by a standard deviation parameter to ensure that the generated failure modes are both diverse and close to the actual scenario.

[0055] The formula for calculating the intensity of fault modes generated by the AI-based multimodal fault simulation algorithm is as follows:

[0056] ;

[0057] in, Indicates the first The intensity of each failure mode; express The combined effects of various fault-influencing factors in this paper ; Indicates the first The fault magnitude of each fault influencing factor is used to measure the magnitude of the impact of the fault influencing factor on the electricity meter fault. It is learned from historical operating data through an artificial intelligence model. This indicates that the transient nature of a fault is simulated using an exponential decay function, ensuring that the impact of the fault does not last indefinitely but gradually weakens over time. Indicates the first The decay rate of each fault influencing factor, controlling the decay speed, reflects the duration of transient faults, and is derived from historical operating data through an artificial intelligence model. The square of the time difference, i.e., the current time Relative to the fault start time The offset is used to calculate the change in fault intensity over time, and the square form ensures that the attenuation is symmetrical; This represents the oscillation characteristics of a periodic fault simulated by a sine function, with the frequency varying from... The phase is determined by Adjustment; Indicates the first The oscillation frequency of each fault influencing factor, the frequency of the control sinusoidal fluctuation, and the speed of oscillation of periodic faults are learned from historical operating data through an artificial intelligence model. Indicates the first The phase shift of each fault influencing factor is used to adjust the starting phase of the sinusoidal fluctuation, determine the initial state of the periodic fault, and avoid complete synchronization of all fault modes. This is learned from historical operating data through an artificial intelligence model. It represents the mutation intensity coefficient, which controls the magnitude of mutative failures, and is derived from historical operational data through an artificial intelligence model; Indicates when The time value is 0, when A time value of 1 simulates the switching characteristics of a sudden fault, ensuring that the fault occurs within... The moment occurs suddenly and persists; The Gaussian noise term is a random variable that follows a normal distribution. By introducing randomness, unpredictable disturbances in the real environment are simulated, making the failure modes closer to actual operating conditions. These are generated by an artificial intelligence model.

[0058] This step utilizes an artificial intelligence model to learn fault characteristics from historical operational data, eliminating the need for manual preset of fault modes and significantly improving the intelligence level of the testing process.

[0059] Step S2: Based on the fault mode intensity, analyze the impact of fault modes on the performance of the electricity meter using a fault mode-based electricity meter performance degradation prediction algorithm, and calculate the response values ​​of the electricity meter performance parameters.

[0060] Specifically, it includes:

[0061] (1) The fault mode-based energy meter performance degradation prediction algorithm introduces a sensitivity coefficient to measure the sensitivity of a specific fault to a specific performance, such as voltage spikes which may have a greater impact on metering accuracy.

[0062] (2) The fault mode-based energy meter performance degradation prediction algorithm also performs nonlinear processing on the fault mode intensity. The index can amplify or weaken the impact of the fault, reflecting the differences in the nonlinear response of different energy meters to the fault.

[0063] (3) The fault mode-based energy meter performance degradation prediction algorithm reflects the fault resistance capability of the performance parameters through the dynamic resistance factor. The resistance baseline under fault-free conditions is used as the initial value. As the intensity of the fault mode increases, the dynamic resistance factor will gradually decrease, reflecting the performance degradation of the energy meter under continuous faults. The decay rate of the dynamic resistance factor is controlled by a degradation coefficient to ensure that the resistance will not drop rapidly due to minor faults. It also integrates the total intensity of all fault modes to ensure that the dynamic adjustment can reflect the superposition effect of multiple faults.

[0064] The formula for calculating the response value of the performance parameters of an electricity meter is as follows:

[0065] ;

[0066] in, Indicates the first The response value of each performance parameter measures the performance of the energy meter under fault conditions. The value ranges from 0 to 1, where 1 indicates no degradation and 0 indicates complete failure. This indicates the cumulative impact of all failure modes on the performance of the electricity meter; The sensitivity coefficient represents the first... The failure mode is related to the first The degree of influence of a performance parameter measures the sensitivity of a specific fault to a specific performance. It can be set according to the specific implementation scenario and is not limited here. Indicates the first The nonlinear transformation of the intensity of each failure mode reflects the nonlinear characteristics of the impact of failure intensity on performance. For example, small failures may have a small impact, while large failures may lead to drastic degradation. Indicates the first The failure mode is related to the first The nonlinear influence index of each performance parameter adjusts the degree of influence of the fault mode intensity, reflecting the differences in the nonlinear response of different individual energy meters to faults. It can be set according to the specific implementation scenario and is not limited here. This represents the resistance decay effect, simulating the decrease in the fault resistance of an energy meter as the fault intensity increases through an exponential function. The exponential form ensures a smooth transition. Avoid having a denominator of zero; This represents the dynamic resistance factor, i.e., the first... The fault tolerance of each performance parameter is calculated using the following formula:

[0067] ;

[0068] in, Indicates the first The baseline of the resistance of each performance parameter under fault-free conditions serves as the initial value of the resistance, reflecting the fault resistance capability of the energy meter under ideal conditions. It can be set according to the specific implementation scenario and is not limited here. Represents the degradation coefficient, controlling the first The coefficient of the resistance decay rate is used to adjust the sensitivity of the resistance to the intensity of the failure mode. It represents the sum of the intensities of all failure modes.

[0069] This step, by taking the fault mode as input, can quickly calculate the degree of performance degradation and generate an intuitive response vector. The automated process reduces the time and cost of manual testing, and is particularly efficient when testing a large number of electricity meters or simulating complex fault scenarios.

[0070] Step S3: Based on the response values ​​of the energy meter performance parameters, calculate the verification error of each performance parameter using a digital energy meter performance verification algorithm.

[0071] The digital energy meter performance verification algorithm processes each performance parameter individually. Specifically:

[0072] (1) Extract the response values ​​of the performance parameters of the electricity meter and the corresponding standard thresholds, calculate the relative deviation between the two, and introduce a pre-set weighting factor to reflect the importance of different performance parameters in the overall performance of the electricity meter. The influence of positive and negative signs is eliminated by squaring to reflect the absolute magnitude of the deviation.

[0073] (2) Calculate a dynamic severity factor based on the cumulative information of the fault mode intensity to enhance the sensitivity of the error to severe faults. Specifically, use an initial severity coefficient as a baseline value, and then dynamically adjust it according to the ratio of the total intensity of all current fault modes to the preset maximum fault mode intensity.

[0074] (3) A dynamic adjustment term is introduced to adjust the verification error based on the severity of the fault and the degree of performance degradation. Specifically, the reciprocal of the response value of the electricity meter performance parameter is taken. To avoid calculation problems caused by the electricity meter performance parameter response value being zero, a very small smoothing constant is added to the electricity meter performance parameter response value, and then multiplied by the dynamic severity factor to obtain an amplification term. The lower the performance prediction value (i.e., the more severe the degradation), the larger the reciprocal. Combined with the dynamic severity factor, the verification error will be significantly amplified, making it more sensitive to capture the negative impact of severe faults on performance.

[0075] (4) In order to make the magnitude of the verification error intuitive and easy to analyze, the square root is performed to obtain the final verification error value of each performance parameter. The square root operation retains the positive value of the error and adjusts the value to a more reasonable range, ensuring that the error is neither exaggerated nor distorted.

[0076] The formula for calculating the performance parameter verification error is:

[0077] ;

[0078] in, Indicates the first The verification error of each performance parameter reflects the degree of deviation between the predicted performance value and the standard value. The squared term representing the relative error, including a weighting factor, calculates the relative deviation between the predicted performance value and the standard value. The squared term eliminates the influence of the positive and negative signs, while the weighting factor is introduced to highlight the importance of the performance. Indicates the first Standard thresholds for each performance parameter; Indicates the first The weighting factor for each performance parameter is used to reflect the importance of different performance parameters. It can be set according to the specific implementation scenario and is not limited here. This indicates a dynamic adjustment item used to adjust the error based on the severity of the fault and the degree of performance degradation. Represents the reciprocal form of the performance degradation sensitive item; This represents a smoothing constant, used to avoid the denominator being zero; Indicates the first The dynamic severity factor of each performance parameter, which dynamically adjusts the sensitivity of the error based on the failure mode intensity, is calculated using the following formula:

[0079] ;

[0080] in, This represents the initial severity coefficient, which serves as the baseline value for the dynamic severity factor. This represents the normalized total fault strength. This represents the preset upper limit of the fault mode intensity, used to standardize the impact of faults. It can be set according to the specific implementation scenario and is not limited here.

[0081] This step, through the design of weighting factors and dynamic severity factors, can adjust the verification error calculation according to the importance of performance parameters and the severity of faults, making the verification results closer to the actual application requirements.

[0082] Step S4: Verify the error based on the performance parameters, and optimize the parameters using an adaptive parameter update algorithm.

[0083] Before adjusting the parameters, it is necessary to determine the direction and extent of the impact of parameter changes on the verification error, i.e., to calculate the parameter sensitivity. Specifically, it is necessary to analyze the sensitivity of the verification error to the fault mode and the sensitivity of the fault mode to the target parameter, and combine them using the chain method to obtain the overall impact of the parameter on the verification error. To improve the convergence speed, an exponential decay factor is introduced in this step. When the error is large, the value of the exponential decay factor is small, and the impact of sensitivity is amplified, thereby speeding up the adjustment. When the error is small, the exponential decay factor is close to 1, and the impact of sensitivity is moderately weakened to avoid over-adjustment.

[0084] Furthermore, to avoid oscillations during the optimization process, a momentum term is introduced to smooth the parameter update process and reduce oscillations during adjustment. The momentum term is calculated based on the difference between the current parameter value and the previous value, multiplied by a momentum coefficient. This helps the adaptive parameter update algorithm find the optimization direction more quickly in the parameter space, especially when the error surface is relatively flat or multiple local extrema exist. In the first iteration, since there is no value from the previous iteration, the momentum term is set to zero, and it gradually takes effect from the second iteration onwards.

[0085] The parameter update formula is as follows:

[0086] ;

[0087] in, Indicates the first In the nth iteration The fault magnitude of each fault mode; Indicates the first In the nth iteration The fault magnitude of each fault mode; Indicates the first In the nth iteration The fault magnitude of each fault mode; This represents the momentum term, used to reduce oscillations during parameter updates and ensure the smoothness of the update process. This represents the momentum coefficient, which is used to accelerate the optimization process and reduce oscillations in parameter updates. It can be set according to the specific implementation scenario and is not limited here. The gradient representing the fault magnitude, in the current iteration, represents the sensitivity of the verification error to the fault magnitude, calculated using the following formula:

[0088] ;

[0089] in, Indicates the first The performance parameter affects the first Sensitivity to each failure mode; Indicates the first The failure magnitude of the first failure mode affects the second failure mode. The impact of each failure mode; This represents the convergence coefficient, which is used to accelerate convergence. It can be set according to the specific implementation scenario and is not limited here. The norm of the error reflects the magnitude of the overall error, and its calculation formula is: .

[0090] To ensure the physical reasonableness of the parameters, a range constraint is set to prevent them from exceeding the practically acceptable range. Specifically, a range of minimum and maximum values ​​is set, determined by the operating characteristics of the electricity meter or the testing standards:

[0091] ;

[0092] in, and These are the minimum and maximum values ​​of the fault magnitude, which can be set according to the specific implementation scenario and are not limited here; This indicates taking the maximum value; This indicates taking the minimum value.

[0093] The parameters for the new round of iterations reflect both the optimization requirements and practical feasibility. By adjusting the parameters to gradually reduce the error, the performance prediction results of the electricity meter are made as close as possible to the standard value. The iteration will continue until the error converges to a preset threshold or the maximum number of iterations is reached.

[0094] The above steps complete an AI-based fault simulation test method for verifying digital energy meters.

[0095] It should be noted that, as will be apparent to those skilled in the art, the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics thereof. The scope of the present invention is defined by the claims rather than the foregoing description.

Claims

1. An AI-based fault simulation analog test digital electric energy meter calibration method, characterized in that it comprises the following steps: Step S1, obtaining the historical operation data of the electric energy meter, using an artificial intelligence-based multi-modal fault simulation algorithm to generate fault modes by analyzing the historical operation data of the electric energy meter; Step S2, based on the fault mode intensity, analyzing the influence of the fault mode on the performance of the electric energy meter through an electric energy meter performance degradation prediction algorithm based on the fault mode, and calculating the response value of the electric energy meter performance parameter; Step S3, based on the response value of the electric energy meter performance parameter, calculating the calibration error of each performance parameter through a digital electric energy meter performance calibration algorithm; In step S3, the digital electric energy meter performance calibration algorithm processes each performance parameter one by one, and the processing method is as follows: Extract the response value of the electric energy meter performance parameter and the corresponding standard threshold value, calculate the relative deviation of the two, introduce a pre-set weighting factor to reflect the importance of different performance parameters in the overall performance of the electric energy meter, and eliminate the influence of positive and negative signs by squaring to reflect the absolute size of the deviation degree; According to the cumulative information of the fault mode intensity, a dynamic severity factor is calculated to enhance the sensitivity of the error to severe faults; Introduce a dynamic adjustment term to adjust the calibration error according to the fault severity and performance degradation degree; the specific method is: take the reciprocal of the sum of the electric energy meter performance parameter response value and the smoothing constant, and then multiply it by the dynamic severity factor to get the amplification term, which amplifies the calibration error to more sensitively capture the negative impact of severe faults on performance; Square the above calculation result to get the final calibration error value of each performance parameter; Step S4, according to the performance parameter calibration error, optimize the parameters through an adaptive parameter updating algorithm. The step S1 specifically comprises:

2. The AI-based fault simulation analog test digital electric energy meter calibration method of claim 1, wherein Step S1-1, obtaining the historical operation data of the electric energy meter, the historical operation data including the running records of the electric energy meter under various environmental conditions, dividing the electric energy meter historical operation data into training set and validation set, the training set being used to train the artificial intelligence model, and the validation set being used to evaluate the accuracy of the fault mode generated by the artificial intelligence model; Step S1-2, using an artificial intelligence-based multi-modal fault simulation algorithm to generate multiple fault modes by analyzing the historical operation data of the electric energy meter. The artificial intelligence-based multi-modal fault simulation algorithm generates fault modes by analyzing the historical operation data of the electric energy meter, combines three types of fault influencing factors: periodic, transient and sudden, and introduces random noise to enhance the authenticity, simulating complex fault scenarios that the digital electric energy meter may encounter in actual operation; 3. The AI-based fault simulation analog test digital electric energy meter calibration method of claim 2, wherein Periodic fault is a disturbance that fluctuates regularly over time, used to simulate a sinusoidal fluctuation mode; Transient fault is simulated by an exponential decay function, which simulates rapid changes in a short period of time, and the decay rate is adjusted to control the duration and impact range of the transient fault; Sudden fault is used to simulate sudden events; Random noise is generated according to normal distribution, and the fluctuation range is controlled by a standard deviation parameter. ​ 4. The AI-based fault simulation analog test digital electric energy meter calibration method of claim 1, wherein: In step S2, the performance degradation prediction algorithm based on the fault mode calculates the direct influence on the specific performance parameters of the electric energy meter for each fault mode, and introduces a sensitivity coefficient to measure the sensitivity of the specific fault to the specific performance.

5. The AI-based fault simulation analog test digital electric energy meter calibration method of claim 4, wherein: In step S2, the performance degradation prediction algorithm based on the fault mode performs nonlinear processing on the fault mode intensity to reflect the nonlinear response difference of different electric energy meter individuals to the fault.

6. The AI-based fault simulation analog test digital electric energy meter calibration method of claim 5, wherein: In step S2, the performance degradation prediction algorithm based on the fault mode reflects the anti-fault ability of the performance parameters through a dynamic resistance factor. The resistance baseline under the condition of no fault is used as the initial value. With the increase of the fault mode intensity, the dynamic resistance factor gradually decreases, reflecting the performance degradation of the electric energy meter under continuous fault. The decay rate of the dynamic resistance factor is controlled by the degradation coefficient.

7. The AI-based fault simulation analog test digital electric energy meter calibration method of claim 1, wherein: When calculating the dynamic severity factor, an initial severity coefficient is used as the reference value, and then the dynamic adjustment is performed according to the ratio of the total intensity of all fault modes to the preset maximum fault mode intensity.

8. The AI-based fault simulation analog test digital electric energy meter calibration method of claim 1, wherein: In step S4, before adjusting the parameters, the influence direction and degree of the change of the parameters on the verification error are determined, that is, the sensitivity of the parameters is calculated: the sensitivity of the verification error to the fault mode and the sensitivity of the fault mode to the target parameter are analyzed, combined together through the chain method, and the overall influence of the parameters on the verification error is obtained; an exponential decay factor is introduced, when the error is large, the value of the exponential decay factor is small, the influence of the sensitivity is amplified, so as to speed up the adjustment, when the error is small, the exponential decay factor is close to 1, the influence of the sensitivity is appropriately weakened; A momentum term is introduced to smooth the updating process of the parameters and reduce the shock in the adjustment; the calculation of the momentum term is based on the difference between the current parameter value and the previous parameter value, and then multiplied by a momentum coefficient.

9. The AI-based fault simulation analog test digital electric energy meter calibration method of claim 8, wherein: In step S4, in order to ensure the physical rationality of the parameters, a range constraint is set to prevent the parameters from exceeding the actual acceptable range.

Images